Syntax and Semantics Babble (S-Babble) is a weekly informal discussion group at UC San Diego on theoretical and experimental aspects of syntax and semantics and related areas such as their interfaces with morphology, pragmatics, and philosophy of language
S-Babble is open to all students and faculty
Speakers are asked to prepare a talk that is no longer than 50 mins and send a title as early as possible, but at least one week before the talk
An abstract of the talk and suggestions for background readings are welcome
Speakers are fully free to choose among the following 3 presentation options
1. Full talk without interruptions except for few brief, specific, crucial clarification questions
2. Beginning of the talk (up to a specified point) without interruptions except for few brief, specific, crucial clarification questions
3. The talk can be interrupted with questions at any time
Tuesday 11 am - 12:30 pm
APM 4218 and/or Zoom (email ivano [at] ucsd [dot] edu to be added to our weekly email list to receive Zoom links)
There are at least three major crosslinguistic generalizations concerning interrogative and non-interrogative wh-clauses as well as their characterizing restricted class of lexical itemsâwh-words. They are supported by findings that have emerged in the past couple of decades, although further data and analysis are needed, especially for non-interrogative wh-clauses. To the best of my knowledge, these generalizations havenât been fully stated together, let alone accounted for. I argue that a principled explanation of these generalizations requires an analysis of all non-interrogative wh-clauses as derived from interrogative wh-clauses and, therefore, sharing a propositional core. This would be the semantic correlate of the shared morphosyntactic core characterizing all wh-clauses. In turn, all wh-words in wh-clauses must be analyzed as logical operators that apply to propositional content. To the best of my knowledge, such a general analysis has not yet been attempted. I formulate and discuss those three generalizations and account for them by developing a âpropositionalâ analysis for each variety of wh-clause and its wh-words.
The theoretical literature on KanienâkĂŠha and related Northern Iroquoian languages has historically been polarized into two broad sides: (i) A âlexicalist / functionalistâ approach (e.g., Mithun 1999, Koenig and Michelson 2012), which assumes a traditional separation between word-level morphology and sentence-level syntax; and (ii) a âgenerativistâ approach (e.g., Baker 1996, Coon in press), which adopts a Distributed Morphology âSingle-Engineâ system in which this traditional morphology-syntax dichotomy disappears. In this talk, I will argue that these two seemingly contradictory views can in fact be easily reconciled by recognizing that, while the theoretical assumptions of the generativist approach are superior, the lexicalist approach provides crucial empirical insights, that must thus be integrated into our analysis; thereby offering a more comprehensive picture of the KanienâkĂŠha language (and potentially other Northern Iroquoian languages).Â
Specifically, we observe that, while word-internal âmorphologyâ is easily amenable to a DM-style syntactic analysis, in which the verbal complex is essentially the realization of a large array of functional projections; the structure of multi-word sequences does not seem to be governed by the same core syntactic principle of selection, in the traditional sense of one object placing constraints on the presence, form, or position of another. In other words, these higher structures seem to be under-determined by the syntax (in the narrow sense of âselection-based structureâ), and may be better characterized as belonging to discourse; but I will argue that we do not need to worry that this is an ad-hoc analysis which abandons well-established assumptions. The formal separation between âinside wordsâ and âoutside wordsâ is thus indeed present in the language, but is reframed as syntax vs. discourse, rather than morphology vs. syntax.Â
To provide a more concrete empirical focus to this rather big-picture talk, I will examine different kinds of discourse constructions which govern the higher discursive structures (in contrast with the âregularâ syntax inside verbs).
Different mental state verbs express different relationships between the attitude holder and the embedded proposition. In contrast to nonfactive belief verbs, such as the English âthinkâ and Mandarin âjuedeâ (to think/feel), the Mandarin contrafactive belief verb âyiweiâ is negatively biased and commonly used to report false beliefs (Glass, 2023, 2025). Previous studies show that children can distinguish between nonfactives and contrafactives by 4 years of age, and the use of âyiweiâ facilitates understanding of false-belief sentences and false-belief abilities more generally (Lee et al., 1999; Zhang & Zhou, 2022). However, it remains unclear how children learn that âyiweiâ is contrafactive. One possible factor is the discourse context in which âyiweiâ occurs: compared to âjuede,â âyiweiâ is more likely to occur in contexts that highlight the falsity of the attributed belief. In this talk, we will first review the arguments for treating âyiweiâ as a contrafactive. Then, we will present an empirical study investigating whether discourse information can serve as a learning cue for acquiring âyiweiâ using a modified version of the Human Simulation Paradigm (Gillette et al., 1999). The results suggest that discourse might be one of the cues facilitating the acquisition of âyiwei,â while other linguistic cues remain to be explored.Â
Many early generative approaches to case assignment took case to be the result of the same type of dependency-forming operation that results in morphological agreement (in Minimalism, the operation Agree). Under this type of approach, a functional head (probe) establishes a dependency with a nominal (goal), resulting in case being assigned to that nominal. But this view has been challenged, and to a large degree abandoned, in recent years in favor of configurational approaches to case assignment. Under configurational case approaches, it is a structural configuration (specifically, c-command) between two nominals that results in case being assigned to one of those nominals via configurational case rules. This type of configurational approach can derive so-called "dependent" case patterns, where the presence of morphological case on one nominal reflects -- or is dependent on -- the presence of another nominal in the structure.
In this talk, I explore crosslinguistic data from multiple dependent case patterns and argue that, even in these types of patterns that are taken to be evidence par excellence for configurational approaches to case assignment, we still see the hallmarks of Agree. Specifically, I demonstrate that case assignment is sensitive to: 1) structural locality to specific functional heads rather than other nominals, 2) c-command between functional heads and nominals rather than between nominals, and 3) hierarchy effects that indicate one-to-many probe-goal mappings. By reducing dependent case assignment to an instance of the independently warranted operation Agree, we are able to eliminate the need for configurational case rules from the grammar, resulting in a more parsimonious theory.
We present the first systematic description of headless relative clauses in Tigrinya (Semitic; Eritrea and Ethiopia; SOV). Drawing on novel fieldwork data, we examine three types of headless relatives, namely light-headed, super-free, and free choice relatives. The data show that, unlike btter-studied languages, Tigrinya forms both headed and headless relative clauses using the same morphosyntactic strategy, which involves the relativizer z(ɨ)- without overt wh-elements. Headless relatives in Tigrinya do not rely on interrogative wh-structures, in contrast to languages such as English. Notably, free relative clauses are not attested. Instead, all headless relative clauses share a structural core with headed relatives, differing only in the presence or absence of nominal and determiner layers. By documenting this pattern, we identify a previously underdescribed strategy in Afro-Asiatic. These findings refine the typology of relativization, provide empirical evidence for cross-linguistic variation in relative clause formation, and establish a foundation for formal analyses of Tigrinya and related languages.
Recent experimental work (Xu & Runner 2022, Lyu & Kaiser 2023) has explored and begun to confirm the often hinted at claim in the theoretical literature (Tang 1989, Pan & Hu 2003, Yu 1992) that taziji in Mandarin is not a reflexive constrained by the traditional Binding Theory, but instead behaves more like a âlong distanceâ (LD) reflexive. Since Mandarin already has the well-known LD reflexive ziji, this raises the question of how exactly taziji and ziji differ, and how these differences can be accounted for theoretically. Much scholarship has focused on the properties of ziji and how to account for them (see Charnavel et al. 2017, Chen 2018 for overviews). In this talk, I will review some of these claims with the goal of designing a study or studies to test the degree to which taziji does or does not have those properties. This is developing work, so I am hoping S-Babble participants will be ready to help brainstorm ideas. Â
In many languages, a morphologically plural pronoun can be interpreted as referentially singular in a coordinate-like structure known as the plural pronoun construction (PPC). In this talk, I describe previously unnoticed PPCs in two Mongolic languages -- Khalkha Mongolian and Kalmyk Oirat -- based on original fieldwork data. Mongolic PPCs exhibit several unusual properties, including (i) the obligatory presence of a numeral, (ii), the absence of exclusory readings, (iii) the absence of third person PPCs, and (iv) in Kalmyk, the use of a morphological subpart of the 1PL pronoun in first person PPCs. Mongolic PPCs thus broaden our understanding of PPCs crosslinguistically and suggest that PPCs do not constitute a uniform typological category.Â
I propose that Mongolic PPCs involve a pronoun containing a CardP projection that selects for an additional DP (the annex). I further argue that first and second person plural forms are conditioned by a plural feature on either Card or Num, while third person plural forms are conditioned only by a plural feature on Num. The proposed analysis supports a decompositional approach to Mongolic pronouns and enriches our understanding of variation in plural morphology.
Old Avestan features a cross-linguistically novel type of free relatives: those based on demonstrative-based pronouns, in contrast to the standard definition of free relatives as being introduced by interrogative-based pronouns (Caponigro 2003). In this talk, I will discuss the significance of this novel type of free relatives for formal analyses of this construction, and propose a new account of free relatives at the syntax-semantics interface. I show that this novel analysis minimizes the need for stipulating silent operators as a result of the unique properties of the Old Avestan free relatives, explains diachronic patterns, and has cross-linguistic support.Â
Word order changes (or scrambling) happen in languages like German, Japanese, Russian. Some theories argue that scrambling is A-bar movement driven by [EPP]. However, scrambling does not seem to have its own semantic/discourse property (e.g., question formation), and grammatical weight is said to play a role in scrambling. In this respect, one might suggest that scrambling is no different from extraposition (perhaps PF-movement) because both operations prepose or postpose words/phrases to the left or right of sentences. To investigate the puzzling nature of scrambling, I will present experimental evidence that scrambling in Korean is island-sensitive to relative clauses and coordinate NPs. However, there was no island sensitivity to complex NPs. Even if I characterized scrambling as "heavy NP"-scrambling in my experiments, scrambling could not escape from island constraints.Â
This presentation will report on data concerning wh-arguments and the placement of wh-phrases in DineĚ Bizaad (DB aka Navajo) as part of ongoing research on DB questions. This presentation will show that DB wh-movement does not require a Q-particle and that wh-arguments are sensitive to islands. The analysis suggests wh-ex situ structure may involve a process not driven by the Q-marker in DineĚ Bizaad.
Languages like German are said to make a morphological distinction between unique and anaphoric definites [1]. Evidence comes from two types of bridging anaphora. In (1), the bridged noun author can be associated with the antecedent noun book because such mentioning introduces the set of individual authors into the discourse.
(1) John bought a book today. The author is French. (Product-producer bridging)
The indexical mechanism plays a role in identifying the specific author among others. On the other hand, the bridged noun steering wheel in (2) is identified as the unique part of car.
(2) John was driving down the street. The steering wheel was cold. (Part-whole bridging)
We typically know that there are no two steering wheels in a car and the index plays no role here. If we consider bare NP languages like Mandarin, the bridged noun in (1) cannot appear as bare NPs because there is no DP structure that can host the index [2]. Because there is no definite article in bare NP languages, product-producer bridging in (1) requires demonstratives that project DPs. By contrast, the bridged noun in (2) should appear as bare NPs because the index is redundant or even banned.
However, it has been also noted that there is significant amount of variation among speakers [3], which does not seem to support the claim that bare NPs are unique definites and demonstrative DPs are anaphoric definites [2]. To tackle this issue, I propose to collect empirical evidence using formal judgment tasks that can quantify varying degrees of acceptability. Importantly, I focus on Korean bridging anaphora because it has been said that Korean makes a three-way distinction of definites, namely, bare NPs as unique definites, demonstrative KU as an anaphoric definite, and demonstrative CE as a deictic definite.
[1] Schwarz, F. (2009). Two types of definites in natural language. PhD dissertation, UMass.
[2] Jenks, P. (2018). Articulated definiteness without articles. Linguistic Inquiry.
[3] Dayal, V., & Jiang, L. J. (2022). The puzzle of anaphoric bare nouns in Mandarin: A counterpoint to index!. Linguistic Inquiry.
Wyandot (endonym Waâżdat) is a polysynthetic Northern Iroquoian language that is thought to be most closely related to Wendat (Kopris 2001, Lukaniec 2018). In all Northern Iroquoian languages bound roots may productively host a large number of affixes. In addition to many optional affixes, all verb roots obligatorily host pronominal prefixes (Michelson 2016). The prefixes encode number and person, as well as inclusivity for 1st person categories and gender for 3rd person categories (Michelson 2016). Pronominal prefixes come in at least three morphologically determined series, sometimes called Agent, Patient and Transitive in the Iroquoian literature. There is a large amount of syncretism of these forms, especially with Transitive prefixes, so that not all logically possible combinations of categorical values are expressible as a distinct pronominal prefix (Michelson 2016).
This talk explores Wyandot data drawn from fieldwork in 1911 and 1912 where the presenterâs relatives were principal consultants, situating the study of Wyandot in a minimalist syntactic framework for the first time. The distribution of pronominal prefixes is examined following the hypotheses of Coon 2025 for KanienâkĂŠha (Mohawk), a related Northern Iroquoian language. Phenomena such as the stative shift and the correspondence (or lack of it) between Agent and Patient series pronominal prefixes and semantic theta-roles are considered.
Coon, Jessica. (2025). âInvisible Inanimates and Problems for Polysynthesis in Northern Iroquoianâ. Ms., McGill University.
Kopris, Craig Alexander. (2001). A grammar and dictionary of Wyandot. State University of New York at Buffalo.
Lukaniec, Megan. (2018). âThe elaboration of verbal structure: Wendat (Huron) verb morphologyâ. PhD thesis, UC Santa Barbara.
Michelson, Karin. (2016). âIroquoian Languages.â Oxford Research Encyclopedia of Linguistics.
The morphosyntactic makeup of X-marking and its similarities to wishes has gained significant attention after von Fintel & Iatridou (2023). Turkish illustrates this clearly, as the conditional morpheme -sA appears both with and without past morphology in X-marked conditionals and in desiderative keĹke âwishâ constructions. However, the ingredients of X-marking in Turkish remain debated. While Soykan (2023) argues that - sA alone constitutes X-marking, von Fintel & Iatridou (2023) propose that Turkish X-marking involves both -sA and past-tense morphology. We address two questions: (i) whether past tense in the antecedent is necessary for X-marking in Turkish, and (ii) whether different morphosyntactic realizations yield distinct interpretations.
We conducted a rating study with native Turkish speakers (N = 43) who evaluated the probability of the antecedent across conditional and wish constructions on PcIbex. The findings reveal that past marking in the consequent is sufficient to trigger a substantial decrease in perceived antecedent probability, while additional past marking in the antecedent has no significant effect. Moreover, across standard diagnostics (i.e., existence presuppositions and countermathematicals), Single-Past and Double-Past marked conditionals pattern alike. Overall, the results challenge the claim that X-marking necessarily involves past morphology in both clauses. Following Ippolito and Mirrazi (2026), we suggest that the optionality of antecedent past may be related to Turkish being a non-SOT language, in which present tense can receive a shifted interpretation.
All Babblers are invited to give short informal presentations of data, ideas, or questions related to syntax and/or semantics to kick things off for the year! We encourage you to use this opportunity for group feedback and discussion on any topics or puzzles that you've been musing over lately. Names of those who would like to present will be drawn randomly, and each presenter will have 5-10 minutes (depending on the number of presenters) for presentation and discussion. You are still welcome and encouraged to attend even if you do not have a topic you would like to present.
Using an array of tests, this paper produces a uniquely comprehensive diagnosis for the logophoric interpretation of wh-moved anaphors. The particular wh-constructions investigated here are argued in the literature to provide evidence that intermediate reconstruction sites satisfy Condition A of the binding theory (Barss 1986; Sportiche 2006; Chomsky 2023). Since these wh-moved anaphors are logophors, expressions that do not adhere to binding theory constraints (at least on the surface, see Charnavel 2019), we may dismiss the purported evidence as irrelevant. This paper uses picture noun phrases, the predominant configuration used in the relevant literature to demonstrate intermediate reconstruction for Condition A satisfaction. My explanation of the presented analysis is based on the observation from Pollard & Sag (1992) that an intervening animate can degrade logophors. Assuming the availability of prolog in picture-NPs, as argued by Charnavel & Bryant (2022), the problem of the intervener is circumvented due to wh-movement to a position where the embedded subject no longer intervenes between prolog and its antecedent. Â
This study investigates how Hungarian-acquiring children understand factivity, focusing on the verbs âtudniâ (âknowâ) and âgondolniâ (âthinkâ). Unlike English, Hungarian is a factivity-alternating language that allows both factive and non-factive uses of âtudniâ, raising questions about how children acquire factivity in such a language. Using a task adapted from prior English-language research, Hungarian children aged 36-54 months heard attitude reports involving negation as clues to the location of a hidden object. Results suggest that by age three to four, some Hungarian children treat âtudniâ as factive, while others do notâa developmental pattern paralleling prior English findings. However, children showed limited understanding of the distinction between âtudniâ and âgondolniâ, suggesting that the factive alternation for âtudniâ may obscure contrasts in meaning between the two verbs. These findings raise the possibility that phenomena like factivity alternations may impact the developmental trajectory for childrenâs acquisition of belief and knowledge verbs across languages.
It is widely accepted that in wh-in-situ languages, there is a relation between the in-situ wh-phrase and its scope position, but accounts differ as to exactly what this relation is. Some have argued that the wh-phrase itself covertly moves to the scope position (e.g., Huang, 1982), others have claimed that the wh-phrase does not move, but that there is still an A-bar dependency between the scope position and the wh-phrase (perhaps mediated by a null operator, as in Aoun & Li, 1993), and still others have said that it is a binding relation rather than an A-bar dependency (e.g., Tsai, 1999). In this paper, we see that these three approaches make different predictions as to the presence/absence of wh-island and D-linking effects in wh-in-situ languages, and by means of a formal sentence acceptability experiment with Cantonese, we show that the predictions of the A-bar dependency account are correct. More specifically, we present evidence that Cantonese exhibits clear wh-island effects, but not D-linking effects, just as would be expected if the relation between the scope position and the in-situ wh-phrase is an A-bar dependency, but without covert movement of the wh-phrase.